Consolidate Llama model usage in tests (#13094)

tests/basic_correctness/test_chunked_prefill.py

@@ -20,7 +20,7 @@ from ..utils import multi_gpu_test
 
 MODELS = [
     "facebook/opt-125m",
-    "meta-llama/Llama-3.2-1B",
+    "meta-llama/Llama-3.2-1B-Instruct",
 ]
 
 
@@ -92,7 +92,7 @@ def test_models_distributed(
 ) -> None:
     override_backend_env_variable(monkeypatch, attention_backend)
 
-    if (model == "meta-llama/Llama-2-7b-hf"
+    if (model == "meta-llama/Llama-3.2-1B-Instruct"
             and distributed_executor_backend == "ray"):
         # test ray adag
         os.environ['VLLM_USE_RAY_SPMD_WORKER'] = "1"
@@ -221,7 +221,7 @@ def test_with_prefix_caching(
     Checks exact match decode with and without prefix caching
     with chunked prefill enabled.
     """
-    model = "meta-llama/Llama-2-7b-chat-hf"
+    model = "meta-llama/Llama-3.2-1B-Instruct"
     # The common prompt has 142 tokens with Llama-2 tokenizer.
     common_prompt = "You are a helpful AI assistant " * 20
     unique_prompts = [